AI applied to Navy condition-based maintenance

Artificial intelligence technologies are maturing rapidly, and federal agencies are looking for opportunities to improve their operations using AI. Our Navy client partners are accumulating more data by volume than ever before. It is our mission to help clients seize the potential of this information for data-driven decision-making. We offer our clients real-time observability and forecasting tailored to their systems, down to the sensor level, in the active and post-mission domains.

This form of system awareness facilitates a long-term system integrity capability for a predictive maintenance implementation. Our strategy incorporates real-time and historical data analyses to provide both sailors and leadership the ability to predict failures ahead of time and rapidly address faults when they arise.

Streamlining data at the dashboard

Our solution provides the sailor with an intuitive and straightforward display during system operation. The platform includes data visualization, device and system forecasting, and collaborative functions that enhance visibility to break down data silos and open communication across the desired team. This approach enables more timely action and less time devoted to maintenance for the system.

In a summary dashboard, we provide the crew with a quick means of identifying the current state of the system and alert on impending issues, reducing the time the sailor needs to evaluate the system status. 

A sample overview of system operation, equipment health feedback, device readings, and system runtime are highlighted in the summary dashboard. Additional levels of analysis are available in supplementary dashboards that include drill-down metrics for a more granular view of device output and trend analysis, where long-term trends allow our client partners to compare metrics like the expected useful lifetime of a device to the actual useful lifetime.

CGI calibrates each visualization to client-specific systems and needs. In addition to the system overview, annotations, alerting, and forecasting are available on the platform.

The AI difference

This is where artificial intelligence, in the form of machine learning, comes in. The alerting functionality is tailored to the system for anomaly and fault detection, while integrated machine learning capabilities predict future states of the system. Example use cases for such a platform design are power systems, pump modules, and monitoring the database hosting the captured data.

One such example is the sinusoidal plot (see figure 2) showing the actual signal versus the predicted signal, days into the future, where the shaded region represents the confidence interval for the prediction. Minute, unexpected fluctuations in a pure signal such as the sine wave, which the user might otherwise overlook, are captured in the small downward deflection of the forecast. Deviation of the predicted signal allows for early inspection of the device, continued monitoring, and, if needed, resolution through actionable steps.

Through this manner of data visualization and AI insights, we equip our clients with the ability to proactively maintain equipment in an informed way by getting ahead of system failures. This, in turn, leads to enhanced safety for the crew, extended useful lifetime of devices in the system by alleviating additional time under malfunctions, increased uptime of the system, and ultimately operational cost savings due to more effective allocation of time and resources to maintenance.

Interested in learning more about this capability? For more on CGI's solutions, visit our website or contact our business development team at navy@cgifederal.com and marinecorps@cgifederal.com.